Meta's seventh Apex PPA buys Texas grid access, not electrons
The 144MW Starling deal keeps Meta's clean-energy claims on ERCOT's grid while the price stays off the record.
Meta has signed a 144MW solar power purchase agreement with Apex Clean Energy in Texas, the seventh deal between the two companies and, as Data Center Dynamics first reported, one that describes the mechanics of hyperscaler clean-energy procurement more plainly than its headline capacity: the power is tied to the Starling Solar project in Gonzales County and delivered to the ERCOT grid rather than to a Meta data center. In exchange, Meta takes exclusive rights to all of the project's environmental attributes, renewable energy credits included, while construction proceeds toward commercial operation in 2027, with financial terms still undisclosed.
The seven agreements Apex and Meta have now signed carry 1.2GW of combined capacity, the report says, while the five it itemizes — 61.6MW in Virginia, 200MW in Texas, 175MW in Illinois, 197MW in Kansas and 224MW in Iowa — sum to just under 858MW. Add Starling's 144MW and the named total lands at roughly a gigawatt, the same figure Urvi Parekh, Meta's head of renewable energy, uses when she says the partnership has brought "a total of one gigawatt of new renewable energy to the grid" across those five states; if the 1.2GW figure for all seven is right, the two agreements the coverage does not name would account for the difference — on the order of 200MW, spread across projects the disclosure leaves off the page.
Because the electrons never reach a Meta building, the transaction is doing two jobs at once: it gives Apex a contracted offtake for a project under construction, and it hands Meta a bundle of environmental attributes it can count against its 100 percent renewable energy commitment without owning or operating generation. Meta's Texas load is real — campuses are running in Fort Worth and Temple, and a third is under development in El Paso for 2028 — but Starling will not serve any of them directly, and that gap is the whole design. Clearing ERCOT's queue and breaking ground is what makes a project financeable, so the resource a hyperscaler is really bidding for is a developer who already holds the permit; this publication has argued that grid permission is the underwriting asset and that queue positions and interconnection contracts price before electrons. Meta's practice of accumulating permitted projects a few hundred megawatts at a time, rather than signing one enormous contract, is that argument seen from the offtake side.
The developer holds the permit
Apex is the counterparty Meta keeps returning to, and its leverage comes from the size of its shelf: the Charlottesville developer says its pipeline exceeds 25GW and that it has commercialized more than 12GW of clean generation since it was founded in 2009, which puts it among the few firms able to hand a hyperscaler a construction-ready Texas project on request. Apex has signed PPAs with other hyperscalers, including Google, according to the report, and that overlap is the tell: when the same developer shows up across the buyer roster, the advantage sits with whoever holds shovel-ready megawatts, which suggests the hyperscalers are competing for permits as much as for power.
Neither company has disclosed what Meta is paying, and the silence is by now routine in these deals, which matters to anyone trying to underwrite the asset class. With no price on the record, the market cannot mark what a hyperscaler offtake is worth, and the merchant risk a fixed-price contract would shift to the buyer stays, on the public record at least, with the developer and its lenders. The blank price column has migrated from projects to fund managers and now rests on the cap table, which makes merchant risk an ownership-level question rather than a developer-financing one; a deal shaped like Starling — attributes to the buyer, output to the grid, price undisclosed — is precisely the instrument that keeps the column blank. One under-construction asset can be financed against a contract like this; whether a 25GW pipeline can be financed against hundreds of them is what the missing number would settle.
Starling is part of a broader Texas pattern: Meta signed a 220MW solar PPA with Sabanci Renewables most recently, and before that a corporate PPA with RWE for the 298MW Rabbit's Foot Solar project in northeast Texas, so the named Texas solar commitments, once the Apex book is added, run to roughly 862MW — every megawatt of it feeding ERCOT and not one of them wired to a campus. For Meta, the arrangement buys clean-energy claims without the cost and delay of building generation behind the meter; for ERCOT, it adds capacity whose output a single buyer has already claimed while the electrons themselves circulate to whoever needs them.
Starling's 2027 commercial operation and the queue behind it are the test — whether Apex keeps producing projects in states where the interconnection and the offtake line up at the same moment — and Meta's El Paso campus, due in 2028, will be counting the attributes it buys today.
With no price on the record, the market cannot mark what a hyperscaler offtake is worth.